A study by Professor Dennis Murphy Odo found that automatic text simplification (ATS) software makes authentic materials more comprehensible for L2 learners with high reading proficiency. However, the tool's effectiveness is limited for learners with lower reading proficiencies. The researcher suggests that ATS tools need to be further...
Researchers found that AI language models exhibit significant implicit bias against individuals with disabilities, resulting in inaccurate sentiment analysis and potential misclassification of posts as toxic. The models' biases were apparent in various applications, including autocorrect and social media content moderation.
Researchers from McGill University and MIT developed an AI system that can learn the rules and patterns of human languages on its own. The model automatically generates higher-level language patterns that can be applied to different languages, achieving better results.
Research finds that certain features of 'baby talk' such as pitch, melody, and articulation rates have consistent properties across most world languages. However, the exaggeration of vowel sound differences varies significantly between languages. As infants develop language skills, IDS gradually becomes more similar to adult speech style.
Researchers found that infants differentiate sounds based on neighboring sounds and vowel durations. Their study suggests that babies can learn the speech contrasts of their language through naturalistic speech data.
A team of researchers at USC has developed a system to quantify and fix anti-queer bias in AI-powered text prediction. By training the popular language model BERT on queer-inclusive data from Twitter and news articles, they significantly reduced biases in predictions, with QueerTwitter showing the most effective results.
Researchers have dated the emergence of genetic variants characteristic of modern humans to two distinct periods: one around 40,000 years ago, linked to population growth and divergence from Africa, and an older one over 100,000 years ago, reflecting the species' greatest diversity in Africa. These findings complement the mosaic nature...
A recent study by ICTA-UAB researchers found that the COVID-19 pandemic has led to a shift in public expectations, with many citizens becoming more pessimistic about future climate action. However, some respondents expressed positive views on the impact of COVID-19 on consumption habits and environmental awareness.
A recent study found that American Sign Language (ASL) features a copula called SELF, previously thought to be absent. Researchers tracked the evolution of SELF through historical records and social media videos, discovering a pattern termed the 'copula cycle' that spans over 150 years
The study found that variability in input leads to better generalization of learned concepts, with relevance being key. Researchers also discovered four types of variability affecting learning, including set size and scheduling, with the 'Mr. Miyagi principle' suggesting unrelated skills can benefit from practice.
The study analyzed over 40,000 words from English texts and found that monosyllabic Middle English words predominantly had long vowels, which were more easily processed by the brain. This suggests that our brains are wired to learn and prefer frequent sound patterns, leading to gradual language changes.
MIT researchers develop ExSum, a framework to formalize explanations of machine-learning models into quantifiable rules. This allows for testing assumptions about model behavior and reveals unexpected insights, such as negative words having sharper contributions to model decisions.
The Uralic language family's history is reevaluated, with research suggesting the languages' westward spread began in western Siberia during the early Metal Age. The dispersal of the family is linked to three episodes: a natural environment shift, cross-cultural archaeological phenomena, and contact with Indo-Iranian languages.
A Cornell study finds that children learn demonstratives like 'this' and 'that' at extremely young ages, even in languages with different structural systems. The research confirms that caregivers can expect children to start using these words between 12-18 months old.
A new study by researchers at New York University found that people tend to prioritize men when referring to individuals in general. The study analyzed over 630 billion words from internet web pages using artificial intelligence tools to measure word meanings.
The Volkswagen Foundation has awarded €1 million to support the development of Sorbian Studies, a small discipline. The project aims to strengthen this field through digitalization and interdisciplinary collaboration, resulting in enhanced language courses, study materials, and digital resources.
A new study by Thomas Pepinsky challenges the idea that language shapes thought and behavior, finding no evidence that linguistic differences affect social and economic outcomes. The study reveals that statistical methodologies used in previous research are prone to producing false positives.
The University of Huddersfield and Vietnam's ULIS are conducting a research collaboration to ensure fairness and inclusivity in English language assessments. The project aims to develop an online toolkit and community of practice for teachers, improving student outcomes across Vietnam.
A recent study found that gender bias in linguistics textbooks has not changed since 1997, with male protagonists appearing twice as often as female ones and being associated with more prominent roles. The study suggests that implicit biases against women are perpetuated in language education.
A study led by Alice Cravotta examined the impact of restricting hand gestures on speech. The researchers found that gesturing did not affect the fluency or acoustic characteristics of speech, but participants produced speech with different prosodic features when gesture production was prevented.
Two studies published in Language reveal persistent gender bias in both peer-reviewed journal articles and course materials used in linguistics. Linguistics textbooks exhibit male-gendered protagonists more often, while research articles show stereotypes about men and women, with non-binary identities absent.
Researchers discovered that the majority of people worldwide exhibit the 'bouba/kiki effect', matching sounds with shapes, regardless of language or writing system. This finding suggests that iconic vocalisations may form a global basis for creating new words.
Researchers from HSE School of Linguistics and Centre for Language and Brain conducted an experiment to understand how language speakers perceive polysemous words. The proximity of figurative senses to the literal sense affects perception as semantically different.
Researchers investigated how people understand emoji by replacing words with emojis in sentences and measuring reading time. Participants easily understood sentences with emoji, but took longer to read emoji depicting homophones, indicating the image must be interpreted first.
A recent study found that bilinguals use the same neural mechanism to combine words from different languages as they do with single-language expressions. This suggests that language switching is natural for bilinguals due to a combinatory mechanism in the brain that doesn't detect language changes.
Researchers found that people's pupils dilate more in response to metaphors than to literal or concrete statements. The study suggests that everyday metaphors, often considered clichés, play a significant role in engaging our brains and conveying richer meaning.
A study led by Ludovic Orlando found that modern domestic horses were first domesticated in the Pontic-Caspian steppes around 2000-2200 BC. The genetic profile of these horses spread rapidly across Eurasia, replacing other horse populations and facilitating their conquest of the region.
Ancient Italian writings suggest that Genoese sailors were aware of North America 150 years before Columbus' famous voyage. The discovery, made by Professor Paolo Chiesa, refers to an area known as 'Markland'/ 'Marckalada', mentioned in Icelandic sources and identified as part of the Atlantic coast of North America.
A team of researchers used a novel approach called onomasiological reconstruction to analyze 5,000-year-old Indo-European and Altaic macrofamily languages. The study confirmed well-documented genealogies and identified relationships between major branches, providing insights into the evolutionary paths of human languages.
An international team reconstructed language families spanning over 10,000 years by combining genetics, linguistics, and musicology data. Grammar was found to reflect population history more closely than other cultural features, highlighting the importance of the grammatical factor in understanding human cultural evolution.
A new study generated 137 whole-genome sequences from eight Middle Eastern populations, filling a major gap in international genomic projects. The researchers found 4.8 million previously undiscovered genetic variants, which could hold medical relevance.
Researchers develop AI system that uses cooperative strategies to improve human-AI collaboration in negotiations. The system, called CaSiNo, is trained on a dataset of negotiation dialogues and improves upon traditional menu-driven interfaces. It has real-world applications in business, education, and entrepreneurship.
Roberta Golinkoff, a renowned University of Delaware professor, has been elected to the National Academy of Education for her groundbreaking contributions to language development and playful learning. Her research focuses on how children learn and grow, with a particular emphasis on the importance of play in early childhood education.
Autistic children who are bilingual show improved theory of mind and executive function skills compared to monolingual peers. Bilingualism enables children to partially compensate for deficits in these areas, allowing them to focus on one language while inhibiting another.
Researchers found that people from diverse linguistic backgrounds could understand novel vocalizations for 30 different meanings, including animate and inanimate entities, actions, properties, and quantifiers. The study suggests iconic vocalisations may have played a crucial role in the creation of original spoken words.
Researchers at TU Darmstadt created an interactive typeface, AdaptiFont, that adjusts font styles to increase reading speed. The system uses machine learning to generate personalized fonts based on individual users' preferences.
A new study profiles teachers providing English Medium Instruction globally, with social sciences and natural sciences being the most represented disciplines. The research also highlights country-specific differences in EMI practices.
A study by University of Oregon researchers found strong support for face coverings on US Twitter, with a majority of users supporting mask-wearing. Media coverage of anti-mask sentiments fueled the rhetoric of those opposed, but the majority of users ignored this rhetoric and instead supported mask-wearing.
Researchers use computer-assisted techniques to reconstruct ancient pronunciation of undocumented words, achieving a success rate of 76% in predicting word forms. The study's findings improve understanding of Western Kho-Bwa languages and their linguistic history.
A Northwestern University study finds that emotional reviews, not star ratings, predict business success in various industries such as movies, books, and restaurants. The research uses computational linguistics to analyze the emotionality of reviews and shows that higher emotionality is associated with greater success.
A study by Toyohashi University of Technology found that Japanese speakers often confuse the pronunciations of /hi/ and /si/ due to palatalized tongue articulation. Real-time MRI and supercomputer simulations revealed that acoustic contrast between the sounds can be produced by differences in tongue shape in the transverse direction.
A UTokyo-MIT study reveals multilingual adults can learn new languages faster and more efficiently than bilinguals, thanks to increased brain activity. Multilingual participants demonstrated superior language acquisition skills and faster processing times in a simulated Kazakh language learning environment.
The study confirms Basques' genetic uniqueness is the result of continuous isolation and scarce gene flow since the Iron Age. The language barrier promoted population separation from subsequent interactions.
A recent study found that bilingual babies have a similar interest in infant-directed speech as monolingual babies, and are developmentally on track. The study showed that exposure to one language does not hinder the learning of another, and that parents can confidently babble to their babies in one or two languages.
A new paper warns of the dangers of large language models, which can perpetuate racist ideologies and cause significant environmental degradation. The models' huge computational needs also restrict access to smaller developers, leaving out marginalized communities.
Researchers found that animals learn to recognize nonsense words with a natural prosodic contour, but struggle with flat contours. This suggests that general principles of language evolution are inherited from other species.
A Universitat Oberta de Catalunya study analyzes 71 Billboard year-end charts from 1946 to 2016 to explore changes in conceptions of love. Metaphors represent over 70% of figurative language used to express feelings of romantic love in popular music.
A new study from Simon Fraser University found that women's voices are underrepresented in Canadian news media, with men quoted three times more often than women. Despite having prominent female leaders, the research team identified key areas for improvement, including hiring more women as reporters.
Researchers investigated how well speech is understood from those speaking while wearing a cloth mask. They found that masked speech was not more difficult to understand and, in some cases, understanding was improved when speakers were instructed to speak clearly. The study highlights the adaptive nature of human speech.
A new study published in the Proceedings of the National Academy of Sciences presents a computationally-based modeling approach that simulates infant language learning. The researchers found that infants do not learn consonant- and vowel-like phonetic categories, but rather learn to distinguish between speech sounds in a more nuanced way.
Researchers developed a powerful tool to map language development and human innovation over thousands of years, including pre-history. The 'linguistic thermometer' uses statistical physics to analyze the evolutionary dynamics of language, revealing how languages change over time.
A University of Miami study found that a soft food diet can reshape language by changing how people speak and pronounce sounds. The research analyzed thousands of languages and the speech patterns of 10 celebrities, including Freddie Mercury and Michael Phelps.
The LSA's 95th Annual Meeting features research on topics like prosodic focus in human- versus voice-AI-directed speech and linguistic measures of symptomatology in schizophrenia. Researchers also explore the bilingual dimensions of [s+] production in Latinx Miami drag culture.
A joint Anglo-German research project will analyze the practice of translation in cultural magazines from 1945-1965. The project aims to understand how these magazines used translated texts to forge new European identities, and explore the significance of translation in reconfiguring a vision of Europe after WWII.
A new Stanford University study identifies 333 Tel Aviv University researchers among the world's top 2%, with 155 in the top 1% and 74 in the top 0.5%. The university is known for its academic excellence and interdisciplinary approach, making this ranking a cause for national pride.
A new study tests machine learning on detecting borrowed words in world languages, showing some promise in specific situations, such as high loanword proportions or single donor languages. However, the results are unsatisfying, highlighting the difficulty of identifying borrowings with mono-lingual information alone.
Researchers found that factual language evokes stronger brain responses than possible language, highlighting the power of fact-based communication in processing information. The study's findings suggest that our brains have a specialized representation for factual information, which is distinct from those for possibilities.
An international team has received a 10 million euro grant to investigate the evolution of human quantification. The QUANTA project will analyze quantification as a bio-cultural phenomenon, combining insights from anthropology, linguistics, archaeology, and cognitive science.
Researchers from Bar-Ilan University demonstrate the application of physical concepts in physics to solve key challenges in artificial intelligence. By adopting power-law scaling, they show that learning each example once is equivalent to learning examples repeatedly, enabling rapid decision-making and ultrafast learning.
A study by RUDN University linguists found that vocabulary size in a foreign language impacts the ability to distinguish between vowel sounds. Russian and English speakers were tested on their ability to match L2 vowels with those of their native tongue, regardless of their vocabulary size.